Announcement

Collapse
No announcement yet.

The Slides Announcing The New "AMDGPU" Kernel Driver

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by bridgman View Post
    We can't make the open driver any more open
    Do i need to remember that statement .

    Today Linux Catalyst has a closed kernel driver and closed X driver but Catalyst/amdgpu will have an open kernel driver and open X driver. How is that not more open ?
    AMD didn't open those parts but will drop some parts in favor and only because that came by part of amdgpu driver design, not by willingless to open anything . Only reasonable thing i see behind amdgpu driver idea is that Catalyst can be used more easely, like that is for opensource existing one .

    Comment


    • Originally posted by dungeon View Post
      AMD didn't open those parts but will drop some parts in favor and only because that came by part of amdgpu driver design, not by willingless to open anything . Only reasonable thing i see behind amdgpu driver idea is that Catalyst can be used more easely, like that is for opensource existing one .
      The main reason for the closed user mode drivers is because they support features today (OpenGL 4.x, OpenCL 2.x, etc.) which are not available today in the open source stack. Moreover, I'm not really sure how useful it would be to open source the catalyst user mode drivers. They use a completely different infrastructure than mesa so they wouldn't really be of value to anyone else outside of AMD and I don't know if there would be much interest in the community in supporting another graphics stack for just AMD hardware when everyone else uses mesa.

      Comment


      • Originally posted by dungeon View Post
        AMD didn't open those parts but will drop some parts in favor and only because that came by part of amdgpu driver design, not by willingless to open anything . Only reasonable thing i see behind amdgpu driver idea is that Catalyst can be used more easely, like that is for opensource existing one .
        Actually we are opening up existing Catalyst code as well by converting key parts from Catalyst to work with the open stack. The current example is Tom puling code from our Catalyst OpenCL intrinsics code and fitting it into libclc for use by the open stack.
        Test signature

        Comment


        • Originally posted by dungeon View Post
          AMD didn't open those parts but will drop some parts in favor and only because that came by part of amdgpu driver design, not by willingless to open anything .
          Hold on, are you saying that if we replace closed source code with equivalent or better open source code that doesn't count as "opening" ?
          Test signature

          Comment


          • Originally posted by bridgman View Post
            Actually we are opening up existing Catalyst code as well by converting key parts from Catalyst to work with the open stack. The current example is Tom puling code from our Catalyst OpenCL intrinsics code and fitting it into libclc for use by the open stack.
            Neat. How platform-dependent is that code? Just wondering id nVidia could finally be convinced to go with OpenCL rather than their own proprietary CUDA

            Comment


            • I'll check, but my recollection is that it's mostly generic code.
              Test signature

              Comment


              • Originally posted by agd5f View Post
                The main reason for the closed user mode drivers is because they support features today (OpenGL 4.x, OpenCL 2.x, etc.) which are not available today in the open source stack.
                Well yeah i know that, i watch your slides and video from XDC. But all this raise bad feeling it will even once matured be overprogrammed bloat or will have unfixable bugs for both, etc. so somewhere in future as a amdgpu driver user and of the opensource one i will feel like second class citizen . who knows

                Moreover, I'm not really sure how useful it would be to open source the catalyst user mode drivers. They use a completely different infrastructure than mesa so they wouldn't really be of value to anyone else outside of AMD and I don't know if there would be much interest in the community in supporting another graphics stack for just AMD hardware when everyone else uses mesa.
                I know, i don't ask to open that just explaing to bridgman that nothing is more open .

                But OK you guys may decide whatever you want/need, as a radeon user simply don't like this unified idea... i can't be optimistic like you and bridgman here, this is a lot controversal

                Comment


                • Originally posted by bridgman View Post
                  Hold on, are you saying that if we replace closed source code with equivalent or better open source code that doesn't count as "opening" ?
                  As prevoius said, i call this something like - unified opensource solution around blobs . Do you think this is something more than that or what existing users does not alredy have with radeon driver?

                  But generaly speaking yes whatever code someone open, for me that means you release it under some of the OSS licences. Generally it does not matter for what purpose it serve, but for me - it does matter .

                  Comment


                  • Originally posted by dungeon View Post
                    As prevoius said, i call this something like - unified opensource solution around blobs .
                    OK, I think this would make a lot more sense for you if you stopped calling it "unified opensource solution around blobs" since it is exactly the opposite

                    Originally posted by dungeon View Post
                    Do you think this is something more than that or what existing users does not alredy have with radeon driver?
                    Yes, it means more developers working on the open source code, ie moving developer focus from closed source to open source for those components. Isn't that exactly what you wanted ?
                    Last edited by bridgman; 13 October 2014, 12:30 PM.
                    Test signature

                    Comment


                    • so bridgeman, thx for some info, so there are coming some features to the free driver because of that unification, thats the "bribe" I talked/asked for.

                      And why would it be bad if the free driver would support opengl 4.0 too? sounded like that no opensource driver user need that

                      Another question would be, if the old chips/cards or the current are not using this new architecture, first is that really the case, I get that there will be a seperate blob with its old kernel part, but will there really be 2 different radeon drm modules for older cards and newer?

                      And if thats really the case would it be really hard to backport some stuff from the new drm module to the old one?

                      Comment

                      Working...
                      X